Mesoscale vortex rapid identification method based on stream function construction

Abstract

The invention relates to a mesoscale vortex rapid identification method based on flow function construction, and the method comprises the steps: inputting the original data of mesoscale vortexes in an identification region into a processing model for data cleaning, and inputting the data into a flow velocity limitation filter and a flow velocity zero-change filter for data preliminary screening; inputting the preliminarily screened vortex information into a flow direction centripetal filter for data re-screening, inputting the re-screened vortex information into a flow function filter, correcting the re-screened vortex information, capturing the shape, obtaining a vortex shape, and calculating the centroid of the vortex shape as a vortex center; and carrying out error comparison on the vortex center and the initial vortex center to obtain confirmed vortex information. According to the method, a main body is constructed in a layered structure based on physical structure characteristics of the vortex, so that rapid fusion of the model is realized, and light weight, easy deployment and easy migration of the model are realized.

Description

technical field [0001] The invention relates to the technical field of marine environment monitoring, in particular to a method for fast identification of mesoscale eddies constructed based on flow functions. Background technique [0002] As a cross-scale ocean dynamic phenomenon, mesoscale eddies have a significant impact on many sea-related activities such as marine fisheries and military operations. In the analysis of marine environmental situation, the identification of mesoscale eddies has important theoretical and practical implications. significance. The existing vortex identification methods, such as data statistics method, curvature center method, neural network method, etc., can be classified into three categories in theory, namely empirical identification and computer learning identification. [0003] Empirical recognition depends on human subjective decision-making, and its recognition efficiency and dynamic adaptation performance are poor; while machine learnin...

AI Technical Summary

Problems solved by technology

[0003] Empirical recognition depends on human subjective decision-making, and the dynamic adaptation performance of its recognition efficiency and correct rate is poor; while machine learning and deep learning recognition based on computer technology, due to the amount of data, high requirements for hardware facilities and long model training time , poor reusability, which limits its application deployment scenarios

Images